ai

9 posts.

• 
  The Evolution of Penetration TestingDay 9 of 100 · AI cybersecurity

  I used to spend most of my time testing web applications. My first look at AI pentesting products suggests the market is moving from point-in-time reports toward continuous, context-rich security testing.

• 
  MCP Attacks on Xata and GitHub MCP: Read-Only Bypass and Issue InjectionDay 8 / 100 · AI cybersecurity

  A first-person walkthrough of two MCP attack labs: a Xata-style read-only SQL bypass and a GitHub public-issue injection that exposed the difference between model safety and application security.

• 
  The MCP confused deputy problemDay 6 / 100 · AI cybersecurity

  A hands-on MCP lab showed how a valid tenant key can still retrieve another tenant's project when the server never checks resource ownership.

• 
  MCP security labs: path bypasses and poisoned tool responsesDay 7 / 100 · AI cybersecurity

  Two MCP labs showed different failures at the same boundary: a filesystem server trusted a string prefix, while a facts server poisoned the agent's next tool call.

• 
  MCP security: when prompt injection gets a toolDay 5 / 100 · AI cybersecurity

  MCP makes it easier for AI systems to use external tools. It also connects model behaviour to permissions, private data, and real actions.

• 
  Poisoning an AI assistant's memoryDay 4 / 100 · AI cybersecurity

  Lakera's MindfulChat challenge made persistent memory feel like a real application attack surface, not just a model prompt.

• 
  An interactive playground for LLM security testingDay 3 / 100 · AI cybersecurity

  I worked through an interactive OWASP LLM Top 10 playground and found a useful starting map, a few strong concepts, and several labs that needed more explanation.

• 
  Web LLM attacks, deeper into indirect prompt injectionDay 2 / 100 · AI cybersecurity

  Indirect prompt injection and insecure output handling showed me that an LLM can turn ordinary application content into an attack path.

• 
  Web LLM attacksDay 1 / 100 · AI cybersecurity

  Mapping an LLM's tools first turned two PortSwigger labs into the same repeatable web-testing method.